Paper: Word Sense Induction using Cluster Ensemble

ACL ID W10-4166
Title Word Sense Induction using Cluster Ensemble
Venue Joint Conference on Chinese Language Processing
Session Main Conference
Year 2010

In this paper, we describe the implementation of an unsupervised learning method for Chinese word sense induction in CIPS-SIGHAN-2010 bakeoff. We present three individual clustering algorithms and the ensemble of them, and discuss in particular different approaches to represent text and select features. Our main system based on cluster ensemble achieves 79.33% in F-score, the best result of this WSI task. Our experiments also demonstrate the versatility and effectiveness of the proposed model on data sparseness problems.